Embed Size (px)
Citation preview
Munich Personal RePEc Archive
Labour Markets, Education and Duality
of Returns
Mamoon, Dawood and Murshed, S. Mansoob
Institute of Social Studies
March 2011
Online at https://mpra.ub.uni-muenchen.de/29529/
MPRA Paper No. 29529, posted 16 Mar 2011 01:18 UTC
1
Labour Markets, Education and Duality of Returns
Dawood Mamoon Pakistan Institute of Trade and Development
State Life Building 7 Islamabad, Pakistan
&
S Mansoob Murshed The Birmingham Business School
University of Birmingham University House, Edgbaston, Birmingham B15 2TT, UK.
Institute of Social Studies (ISS) PO Box 29776
2502 LT, The Hague, The Netherlands Centre for the Study of Civil War (CSCW)
PRIO, Oslo, Norway [email protected]
ABSTRACT
The aim of this paper is to examine the impact of increased trade on wage inequality in developing countries, and whether a higher human capital stock moderates this effect. We look at the skilled-unskilled wage differential. When better educated societies open up their economies, increased trade is likely to induce less inequality on impact because the supply of skills better matches demand. But greater international exposure also brings about technological diffusion, further raising skilled labour demand. This may raise wage inequality, in contrast to the initial egalitarian level effect of human capital. We attempt to
measure these two opposing forces. We also employ a broad set of indicators to measure
trade liberalization policies as well as general openness, which is an outcome, and not a policy variable. We further examine what type of education most reduces inequality. Our findings suggest that countries with a higher level of initial human capital do well on the inequality front, but human capital which accrues through the trade liberalization channel has inegalitarian effects. Our results also have implications for the speed at which trade policies are liberalized, the implication being that better educated nations should liberalize faster.
Keywords: Integration, Trade Liberalization, Wage Inequality.
JEL Classifications: F-15, I-3
2
1 Introduction
Increased trade, particularly of the inter-industry variety, alters the composition of output in the economy away from non-traded goods towards traded products. This will affect the functional distribution of income, usually raising the demand for the factor of production employed intensively in the traded sector. In the developed world, it is skilled labour and we have witnessed an increase in the skilled-unskilled labor relative wage premium. In many OECD countries, this has meant a more unequal personal distribution of income. As far as developing countries are concerned, especially in those that export unskilled labour intensive manufactured goods, we would expect a fall in the skilled-unskilled labour relative wage premium leading to reduced inequality, since the unskilled are more numerous within the population. Yet this is generally not true, and inequality in the developing world has risen, mirroring events in the OECD. What accounts for this paradox? Perhaps developing countries have such quantities of unskilled labour that unskilled wages will not respond to increased demand. This certainly appears likely in cross-country studies where China and India are included. Alternatively, other less populous developing countries may be exporting relatively more skilled labour intensive products such as semi-conductors or capital-intensive commodities as is the case with fuels and minerals. Finally, an expansion in international trade may raise the demand for, and reward of, skilled labour even when the country in question is exporting unskilled labour intensive products due to skill shortages and other factor complementarities.
Many studies have tried to capture the relationship between trade liberalization and
income inequality. A paper by Dollar and Kraay (2004) concludes that liberalization does not significantly affect the distribution of income, and at most, the relationship is of neutral nature. However, their results have been widely challenged because of their methodology and variable choice. (Ravallion 2003; Amann et al. 2002) Ravallion (2003) points out that increased openness can lead to a rise in the demand for relatively skilled labour, which tends towards less equal distribution in poor relative to rich countries. Arbache, Dickerson and Green (2004) find that imported technology raised the relative demand for highly skilled labour in Brazil and thus lowered the relative wages of less educated groups. Behrman, Birdsall and Szekely (2001) observe that inequality has increased in seven out of 18 Latin American countries that initiated market reforms in the mid-1980s. Jayasuriya (2002) accepts that trade liberalization may have reduced consumption poverty in South Asia, but is skeptical about the purportedly neutral distributional effects of liberalization. Many suggest that the distribution of the positive effects of liberalization is somewhat skewed towards urban households rather than rural ones, and to wealthy rather than poor households (see Chen and Ravallion 2003; Cockburn 2002; Friedman 2000; Lofgren 1999). The evidence in this regard comes mainly from Latin America because most of the economies there undertook rigorous reform policies in the mid-1980s following
3
the debt crisis in that decade. Legovini, Bouillon and Lustig (2001) find that inequality in Mexico rose sharply between 1984 and 1994, and rising returns to skilled labour accounted for 20 per cent of the increase in the inequality in household income. Similarly, Hanson and Harrison (1999) find that the reduction in tariffs and the elimination of import licenses accounts for 23 per cent of the increase in the relative wages of skilled labour during 1986-90, thus providing evidence for the role liberalization played in rising inequality in Mexico. Other country studies on Brazil, Chile, Colombia and Venezuela, also show that skilled workers received increased premiums after liberalization when compared to their unskilled counterparts. (World Bank 2001) Therefore, the balance of the evidence points to increased globalization inducing greater income inequality.
In (developing) countries, trade liberalisation is likely to have had conflicting effects on the distribution of earnings. On the one hand, standard trade theory implies that trade liberalisation should result in a reduction in the skill premium. On the other hand, trade flows bring in new technologies and ideas that enhance the productivity of all workers, but especially that of skilled workers. Clearly, the notion of “skilled” and “unskilled” workers differs across countries. In less developed economies, those at the top of the earning distribution often have no more than a secondary degree. Yet, they have skills that will be enhanced by the arrival of new technologies, thus increasing their wage relative to that of uneducated workers.
Irrespective of the exact nature of the cause of trade-induced inequality, it is
sensible to presume that nations with higher stocks of human capital will experience less of the un-equalizing spiral consequent upon globalization and trade liberalization. Investment in education may yield a double dividend. It can not only promote growth, but also suppresses inequality by both bequeathing skills as well as moderating rises in skill-premia following an expansion of international trade. More generally, Tinbergen (1975) pointed out that changes in wage inequality are a result of the opposing forces that technological change (skilled labour demand) and education (skilled labour supply/ human capital) exert on relative wages. Eiche and Garcia-Penalosa (2001) suggest that human capital accumulation plays a dual role in development.
The aim of this paper is to examine the impact of increased trade on inequality,
and investigate whether a higher human capital stock moderates this unequalising aspect of international trade; specifically the skilled-unskilled wage differential. High initial endowments of human capital, captured by data on average years of schooling for example, imply a more egalitarian society compared to countries with a lower human capital endowment. When societies that are more equal, open up their economies further, increased trade is likely to induce less inequality because the supply of skills better matches demand. Yet greater international exposure also brings about technological diffusion, see Winters (2004), further raising skilled labour
4
demand. This may raise wage inequality, in contrast to the initial egalitarian level effect of human capital. This paper will attempt to measure these two opposing forces. An innovation of this paper is to employ a broad set of openness indicators to measure trade-liberalization policies as well as trade policy variables. Another purpose of this analysis is to examine what type of education most reduces inequality. In settings of low human capital endowments, as measured by literacy or low primary school enrolment, a policy of relative neglect of primary in favour of expenditure on tertiary education may have a less than benign influence on inequality. The sample countries here exclude developed nations and economies in transition because of higher stocks of human capital in those regions (see table 1 at the end of the paper). The rest of the paper is as follows, section 2 presents the data and methodology, section 3 contains the empirical results, and finally section 4 concludes with some policy implications.
2 Data and Methodology
2.1 Data Description
We are interested to look into the patterns of wage inequality in growth promoting Industrial sectors of developing countries. The UTIP-UNIDO wage inequality Theil measure which is calculated by University of Texas Inequality Project (UTIP) measures the dispersion of pay across industrial categories in the manufacturing sector. The data on wages is drawn from the Industrial database published annually by United Nations Industrial Development Organization (UNIDO). Our principal reason for using the UTIP-UNIDO wage inequality Theil measure is that we are more interested in the functional distribution of income. Changes in the functional distribution between skilled and unskilled labour, will in turn affect the personal income distribution in countries that are unskilled labour abundant. Inequality will rise in developing countries as the skilled-unskilled labour wage premium increases and vice versa.
5
Table 2
Summary Statistics
Variables Code Source Obs Std . Dev
Dependent
UTIP-UNIDO Wage Inequality THEIL Measure: Developing
countries only, 1999
Theil99de
v
University of Texas
Inequality Project (UTIP) http://utip.gov.utexas.edu
109 (0.108)
Endogenous Independent
Openness Variables
(Exports +Imports)/GDP, 1985 Lcopen Penn World Tables ,
Mark 6
170 (0.589)
Import Penetration: overall, 1985 Impnov85 Pritchett (1996) 96 (21.08)
Import Penetration: manufacturing, 1985 Impnov85
m
Pritchett (1996) 96 (12.79)
Import Penetration: agriculture, 1985 Impnov85
a
Pritchett (1996) 96 (3.818)
Import Penetration: resources, 1985 Impnov85
r
Pritchett (1996) 96 (6.594)
Import Penetration: overall, 1982 Impnov82 Pritchett (1996) 95 (23.85)
Import Penetration: manufacturing, 1982 Impnov82
m
Pritchett (1996) 95 (13.107)
Import Penetration: agriculture, 1982 Impnov82
a
Pritchett (1996) 95 (3.67)
Import Penetration: resources, 1982 Impnov82
r
Pritchett (1996) 95 (9.66)
TARS trade penetration: overall, 1985 Tars85 Pritchett (1996) 96 (36.91)
TARS trade penetration: manufacturing , 1985 Tars85m Pritchett (1996) 96 (21.852)
TARS trade penetration: agriculture, 1985 Tars85a Pritchett (1996) 96 (8.758)
TARS trade penetration: resources, 1985 Tars85r Pritchett (1996) 96 (15.636)
TARS trade penetration,: overall, 1982 Tars82 Pritchett (1996) 93 (83.10)
TARS trade penetration: manufacturing , 1982 Tars82m Pritchett (1996) 93 (26.47)
TARS trade penetration: agriculture, 1982 Tars82a Pritchett (1996) 93 (9.786)
TARS trade penetration: resources, 1982 Tars82r Pritchett (1996) 93 (54.652)
Trade Policy Variables
Tariffs
Import duties as % imports,1985 Tariffs World Development
Indicators
99 (8.903)
Tariffs on international inputs and capital goods, 1985 Owti Sachs and Warner (1995) 98 (0.165
Trade taxes/ trade, 1982 Txtrdg Edwards (1997) 54 (0.031)
Weighted average of total import charges: overall, 1985 Totimpv8
5
Pritchett (1996) 76 (21.30)
Weighted average of total import charges: manufacturing,
1985
Totimpv8
5m
Pritchett (1996) 76 (22.75)
Weighted average of total import charges: agriculture, 1985 Totimpv8
5a
Pritchett (1996) 76 (21.57)
Weighted average of total import charges: agriculture, 1985 Totimpv8
5r
Pritchett (1996) 76 (18.15)
Non Tariff Barriers
Non trade barriers frequency on intermediate inputs, 1985 Owqi Sachs and Warner (1995) 96 (0.242)
Non-tariff barriers Coverage: overall, 1985 Nontar85 Pritchett (1996) 76 (36.305)
Non-tariff barriers Coverage: manufacturing, 1985 Nontar85
m
Pritchett (1996) 76 (37.914)
Non-tariff barriers Coverage: agriculture, 1985 Nontar85
a
Pritchett (1996) 76 (35.268)
Non-tariff barriers Coverage: resources, 1985 Nontar85
r
Pritchett (1996) 76 (43.111)
Composite Measures
Sachs and Warner 1980s Open80s Edwards (1998) 61 (0.446)
Measures based on residuals
Leamers Measure, 1982 Leamer82 Edwards (1997) 47 (0.527)
Gravity-residuals, basic model, 1982 Grmb Hiscos and Kastner 77 (9.922)
6
(2002)
Gravity-residuals, augmented model, 1982 Grma Hiscos and Kastner
(2002)
77 (9.341)
Price Based Measures
Black market premium, 1985 Black Haririson (1996) 61 (0.604)
Skills
Average years of schooling, 1999 Skills99 Baro and Lee (2001) 109 (2.914)
Higher Skills
Average years of higher schooling, 1999 Hyr99 Baro and Lee (2001) 102 (2.914)
Independent :
Initial Skills
Average years of schooling, 1985 Skills85 Baro and Lee (2001) 105 (2.799)
Average years of schooling, 1980 Skills80 Baro and Lee (2001) 105 (2.861)
Average years of schooling, 1975 Skills75 Baro and Lee(2001) 106 (2.703)
Average years of schooling, 1970 Skills70 Baro and Lee (2001) 101 (2.701)
Average years of schooling, 1965 Skills65 Baro and Lee (2001) 99 (2.516)
Average years of schooling, 1960 Skills60 Baro and Lee (2001) 99 (2.522)
Instruments
Natural logarithm of predicted trade shares computed from a
bilateral trade equation with ‘pure geography’ variables, 1999
Lfrkrom Frankel and Romer
(1999)
163 (16.75)
Drop out rate, 1980s Drop80 Barro and Lee (1996) 125 (0.802)
Number of school days Schday Barro and Lee (1996) 139 (23.43)
Distance from the equator of capital city measured as abs
(Latitude)/90
Disteq Acemogolu (2001) 208 (16.65)
The UTIP- UNIDO wage inequality measure is the between-group component of Theil’s T statistic, an entropy measure whose functional form is defined as:
+=+= BW
i
i
ii
i TT
NN
YY
Y
YT
Y
YT )log()( (1)
Where wT and BT indicate within-group and between-group inequality measures
respectively. N and Y stand for total employment and total pay respectively, and
subscript i denotes group identity. As mentioned, UTIP captures BT as their
inequality measure, where groups are defined as categories within the UNIDO industrial classification codes.
The UTIP data set provides Theil inequality measures for nearly 3,200
country/year observations, covering more than 150 countries during the period 1963 to 1999. Figure 1 illustrates trends in wage inequality between skilled and unskilled workers, over time in selected developing countries and is representative of different regions. All the country graphs, except one, show that wage inequality has been on the rise in the 1980s and 1990s. The only exception is Singapore, which belongs to a group associated with the ‘East Asian Miracle’ of the 1980s. However, this miracle remained confined to a few countries and as it is evident from the graphs, Singapore is not representative of the developing world. Since the 1980s and 1990s are associated with Structural Adjustment Policies under which many developing countries embraced trade liberalization, it is safe to suggest that the above trends in wage inequality also relate to these market reforms. The end of the paper lists all
7
India
0
0.02
0.04
0.06
0.08
0.1
0.12
1963
19651967
1969197
119
731975
19771979
1981
19831985
19871989
1991
1993
19951997
1999
Years
Wa
ge
In
eq
ua
lity
Pakistan
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
Years
Wa
ge
In
eq
ua
lity
Zimbabwe
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
1963
19651967
1969
19711973
1975
1977
19791981
1983
19851987
1989
19911993
1995
1997
Year
Wa
ge
In
eq
ua
lity
Bangladesh
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
1967196
91971
1973
197519
771979
19811983
1985198
71989
1991
Years
Wa
ge
In
eq
ua
lity
Singapore
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
Years
Wa
ge
In
eq
ua
lity
Jamaica
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
1963
19651967
1969
19711973
1975
19771979
1981
1983
19851987
1989
1991
Year
Wa
ge
In
eq
ua
lity
China
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0.0035
0.004
0.0045
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986
Years
Wa
ge
In
eq
ua
lity
developing countries, and the latest year for which the Theil wage inequality index is available.
Figure 1 Wage Inequality Trends in Some Selected Developing Countries
Here, we want to capture the effect of education (skilled) versus no education
(unskilled) on relative wages. We would also analyze effect of higher skills within the framework to check if wage inequality between skilled and unskilled labour are rising also because of returns to higher education are rising at higher proportion when compared with overall levels of education. In other words, is wage inequality also pushed by favouring higher skills in developing countries, or presence of skills (having education) a factor decisive enough to explain rise in wage gaps between skilled labour and unskilled labour? There is already some evidence that secondary education is more important in alleviating wage inequality than higher levels of education suggesting close correlation between higher levels of education and wage dispersion (Acemoglu, 2001). Investing in higher education alone is less effective in alleviating wage inequality. Since Theil captures wage inequality and not wage equality, we can easily test the positive effect of higher education in wage inequality. We are not saying that decreasing higher levels of education would then decrease wage inequality as is generally true with interpretations upon getting a positive sign (say between Theil Index and higher levels of education). If there is a positive correlation, then the only way to minimize the education bias of inequality is to raise
Colombia
0
0.01
0.02
0.03
0.04
0.05
0.06
19631965
19671969
19711973
19751977
19791981
19831985
19871989
19911993
19951997
1999
Years
Wa
ge
In
eq
ua
lity
Chile
0
0.02
0.04
0.06
0.08
0.1
0.12
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
Years
Wa
ge
In
eq
ua
lity
8
the overall education levels of the population, which in turn would distribute skills homogenously within the population.
A simple Heckscher-Ohlin or Stolper-Samuelson model would suggest that the
overall return to skills would decline, and with it incentives for education, when a skill-scarce developing country opens up. (Wood and Ridao-Cano 1999) However in a multi-dimensional Stolper-Samuelson model approximating reality, endogenous growth with increasing returns to R & D, a skill-bias in tradables, skill shortages or unlimited supplies of unskilled labour could all lead to an increase in returns to skill following greater integration. (Arbache et al. 2004) Integration can also lead to the diffusion of more efficient education technologies, which would further augment the level of skills in the economy. (Winters 2004) The expected effect of openness and trade policy on wage inequality is positive in developing countries where a majority of the population is unskilled and uneducated.
Initial education levels in a country capture the relative factor supply effect. Trade
on the other hand captures the relative factor demand effect through skill bias technical change as also suggested by Acemoglu (1999; 2001). Other than the initial levels of education, trade becomes the second variable of interest. In view of this, the basic model for wage inequality between skilled and unskilled workers, based on integration is as follows
],([ 0SkillsnIntegrationIntegratioflityWageInequa = (2)
Skills represent education levels. Average years of schooling are a measure of
average skill levels in a country. The initial skills 0Skills in this paper represent
average years of schooling obtained before 1980s. There are outcome based measures (openness) and incidence based (trade policy) measures of integration (trade liberalization). The literature often ignores the latter effect, which only employs openness indicators. Take note however, that openness is an outcome of trade and industrial policies and not a policy indicator per se. Our formal empirical model would have 2 separate identifications then:
]),([ 0SkillsOpennessnIntegratiofInequality = (3)
(+) (-)
]),([ 0SkillsyTradePolicnIntegratiofInequality = (4)
(+) (-)
Here wage inequality is a positive function of integration, which in turn relates positively to the degree of the openness of the economy in Eq. (3), or trade policies that promote greater openness in Eq. (4). Wage inequality negatively relates to the initial stock of skills in the economy as discussed.
9
The econometric form of the wage inequality model based on openness and trade policy is as follows:
iiiiskillsOPENTHEIL 11111 65 ενκσ +++= (5)
iiiiskillsTPTHEIL 22222 65 ενκσ +++= (6)
Where iTHEIL is wage inequality in a country i for the 1990s (employing the latest
value available for the Theil index for every country: see table 1, end of the paper for
the exact year), iε is the random error term,
iOPEN captures openness and iTP is the
indicator for the trade policy stance in the 1980s respectively. Also iskill65 measures
initial skill levels proxied by average years of schooling for the population aged 25 in
1965. Note that the skill acquisition parameter refers to a period well before the trade
liberalization episodes post-1980. This analysis has a specific focus on trade and education. It employs 34 measures
of openness and trade policy to carry out multiple regression analysis for Eqs (5) and (6), respectively. Average years of schooling for 1960, 1970, 1975 and 1985 are also employed for further robustness checks of the model with repeated specifications.
Below, the author presents the taxonomy for outcome based and incidence based
measures of trade following the grouping offered by Rose (2004:):
1. openness (e.g. the ratio of trade or imports to GDP), an outcome based measure,
2. trade flows, adjusted for country-characteristics (outcome based),
3. tariffs (policy incidence-based)
4. non-tariff barriers (NTBs) (incidence based),
5. informal or qualitative measures,
6. composite indices, and,
7. Measures based on price outcomes.
The core openness variable remains the overall trade share (the ratio of nominal
imports plus exports to GDP), which has been extensively used in the literature. (Frankel and Romer 1999; Acemoglu, Johnson and Robinson 2004; Alcala and Ciccone 2002; Dollar and Kraay 2002; Rodrik et al. 2004) Two other measures of openness are trade penetration (TARS) derived from the World Bank’s TARS system and overall import penetration (Impenov) respectively. Pritchett (1996) uses 16 cross-sectional measures of trade penetration for developing countries. These are provided for two different years (1982 and 1985) and four different categories (overall, manufacturing, agriculture and resources sectors). Trade penetration measures are available for imports alone. See table 2 for yearly and sectoral decomposition of these variables.
10
There are many indicators of trade restrictiveness (incidence based) acting as
measures of trade policy (Edwards 1998; Greenaway et al. 2001; Rose 2004): Rodriguez and Rodrik (2000) recommend using simple averages of taxes on imports and exports. A simple import duty as a percentage of imports (Tariffs) is taken from World Development Indicators (WDI). Sachs and Warner (1995) constructed a composite measure of openness by using tariffs on intermediate inputs and capital goods (Owti). Edwards (1998) collected data on total revenues from taxes on international trade as a proportion of total trade (Txtrdg). Pritchett (1996) provides weighted average of total import charges (Totimpov), as well as sectoral categories of import charges (manufacturing, agriculture and resources). They can all be considered good proxies of trade restrictiveness and have been employed in the analysis.
In the category of non tariff barriers, the coverage in terms of total imports is a
widely used measure of trade policy (Rose 2004). Sachs and Warner (1995) recommend the use of frequency of non trade barriers on intermediate inputs (Owqi). Pritchett (1996) provides data on non-tariff barrier coverage for developing countries based on sectoral categories from UNCTAD (United Nations Conference on Trade and Development). The four different categories are manufacturing, agriculture and resources respectively.
Sachs and Warner (1995) develop a composite measure of openness (Open80) but
the measure has a drawback: ‘It provides only a binary classification –a country is open or closed. As are results, countries with different degrees of trade intervention are equally classified as open.’ (Edwards 1998; 385) Open80 is also used in this analysis. Leamer (1988) used the differences between predicted and actual trade intensity ratios as
indicators of trade barriers (Leamer82). A less structural approach is taken by Hiscox and Kastner (2002). They use fixed country-year residual effects from two gravity models of trade (Grmb and Grma) (a simple version which links imports to GDP and distance, and an augmented one which adds measures of wealth, land and capital) to
derive measures of trade policy orientation (Rose 2004) Finally, Sachs and Warner (1995) and Harrison (1996) have utilized a number of
price-based measures of trade policy. The black market foreign exchange premium (Black) is one of them and we have employed it in our study.
Overall, we have been able to obtain 34 measures of integration. These 34
measures of integration would separately enter either equations (5) or (6) along with initial skills to determine wage inequality in at least 34 different regression equations. There could be potential endogeneity problems associated with the dependent variable, wage inequality and the explanatory variable, openness/trade policy. First, openness when measured by the trade share of national income is not truly exogenous, but an outcome of other factors. Second, the degree of wage inequality between skilled and unskilled workers, or the country’s relative factor endowments
11
AGO
ARG
BDI
BEN
BFA
BGD
BHR
BHS
BLZ
BOL
BRA
BRB
BTNBWA
CAF
CHL
CHN
CIV
CMR
COG
COL
CPVCRI
CUB
CYP
DOMDZA
ECUEGY
ETH
FJI GAB
GHA
GMBGNQ
GTM
HKG
HND
HTIIDN
IND
IRN
JAMJOR
KEN
KOR
KWT
LKA
LSO
MAR
MDG
MEX
MMR
MNG
MOZ
MRTMUS
MWI
MYSNAM
NGA
NIC
NPL
OMN
PAK
PAN
PER
PHL
PNG
PRI
PRY
QAT
RWA
SAU
SDN
SEN
SGP
SLV
SUR
SWZ
SYC
SYR
TGO
THA
TTO
TUN
TUR
TWN
TZA UGAURY
VCT
VENYEMZAF
ZWE
.05
.1.1
5
Wage inequalit
y (
Theil
index)
2.5 3 3.5 4 4.5 5Openness (predicted lcopen)
Fitted values Fitted values
AGO
ARG
BDI
BEN
BFA
BGD
BHR
BHS
BLZ
BOL
BRA
BRB
BTNBWA
CAF
CHL
CHN
CIV
CMR
COG
COL
CPVCRI
CUB
CYP
DOMDZA
ECUEGY
ETH
FJIGAB
GHA
GMBGNQ
GTM
HKG
HND
HTI IDN
IND
IRN
JAMJOR
KEN
KOR
KWT
LKA
LSO
MAR
MDG
MEX
MMR
MNG
MOZ
MRTMUS
MWI
MYSNAM
NGA
NIC
NPL
OMN
PAK
PAN
PER
PHL
PNG
PRI
PRY
QAT
RWA
SAU
SDN
SEN
SGP
SLV
SUR
SWZ
SYC
SYR
TGO
THA
TTO
TUN
TUR
TWN
TZAUGA URY
VCT
VENYEM ZAF
ZWE
.05
.1.1
5
Wage inequalit
y (
Theil
index)
.1 .2 .3 .4trade policy (predicted owqi)
Fitted values Fitted values
Afghanistan
AngolaUnited Arab Emirates
Argentina
Azerbaijan
BurundiBeninBurkina Faso
Bangladesh
Bulgaria
BahrainBolivia
BrazilBhutan
Botswana
Central African RepublicChile
China
Cote d'Ivoire
Cameroon
Congo, Rep.Cape Verde
Costa Rica CyprusDominican Republic
AlgeriaEcuadorEgypt, Arab Rep.Eritrea
EthiopiaFiji Gabon
Georgia
Equatorial Guinea
Hong Kong, China
HondurasCroatia Haiti
India
Iraq
JamaicaJordanKenya
Korea, Rep.
Sri Lanka
LesothoLithuaniaLatvia
Macao, China
Moldova
Myanmar
Mozambique
Mauritius
MalawiNigeria NicaraguaNepal
OmanPakistanPanama
PeruPapua New GuineaRomania
Rwanda
Saudi ArabiaSudan
SenegalSurinameSlovak RepublicSlovenia Swaziland
SeychellesTogo
Tonga
Trinidad and Tobago
Tunisia
Turkey
Tanzania
UgandaUkraineUruguay
Uzbekistan
St. Vincent and the Grenadines
Samoa
Yugoslavia, Fed. Rep.Zambia
0.2
.4.6
.8
Wage Inequalit
y
2 3 4 5 6Openness
Theil97 Fitted values
Afghanistan
United Arab Emirates
Azerbaijan
Burundi BeninBurkina Faso
Bulgaria
BahrainBoliviaBrazil
Botswana
Central African RepublicChile
Cote d'Ivoire
Cameroon
Congo, Rep.Cape VerdeDominican Republic
AlgeriaEcuador
EritreaEthiopiaFijiGabon
Georgia
Equatorial GuineaCroatia
India JamaicaJordan
Korea, Rep.
Lithuania
Macao, China
Moldova
Mauritius
MalawiNigeriaNicaraguaNepal
OmanPanama
Peru
Rwanda
Saudi ArabiaSudan
SenegalSlovenia
SeychellesTogo
Tonga
Trinidad and Tobago
Tunisia
Turkey
Tanzania
UkraineUruguay
Uzbekistan
St. Vincent and the Grenadines
Samoa
Zambia
0.2
.4.6
.8
Wage Inequalit
y
0 .5 1 1.5trade policy(tariff rates)
Theil97 Fitted values
(Tavares 1998) may determine a country’s trade policy choices. Even though gradual and under the aegis of structural adjustment policies, decisions regarding openness were taken, it may be that more egalitarian labour abundant nations may choose to open up faster than less equal land (or mineral) abundant nations.
An instrument is required to solve for the openness and trade policy variables’
potential endogeneity with wage inequality. The paper uses the predicted trade share from a gravity equation following Frankel and Romer (1999) to instrument for openness/trade policy. Following Rodrik et al. (2004), distance from the equator is the second instrument for openness/trade policy variables because the level of integration of an economy also depends upon its location in the world map. The Instrumental Variable (IV) regression model is a two stage least squares (2SLS) estimate beginning with
iiii DisteqFROPEN 31111 εψτς +++= (7)
iiii DisteqFRTP 42221 εψτς +++= (8)
Here iFR stands for predicted trade shares from gravity equations computed by
Frankel and Romer (1999), whereasiDisteq (distance from the equator) is a proxy for
geography.
Figure 2 Responsiveness of Wage Inequality w.r.t Openness and Trade Policy
(Actual) (Predicted)
In the first stage, Eqs (7) and (8) generate predicted values of openness and trade policy variables by regressing them on the two instruments. The second and final stage of the regression analysis involves employing the predicted openness and trade
12
policy variables in Eqs (5) and (6) respectively. Before carrying out the IV analysis, it is informative to look at simple bi-variate relations between integration (predicted trade shares and predicted tariff rates) with the Theil index. First column graphs are between openness, trade policy and Theil index for developing countries (Theil99dev) only. Trade indices show no clear relationship with wage inequality. However, once predicted, clear relations develop between trade and wage inequality as the second column in figure 2 shows. Increases in trade shares after liberalization leads to higher wage inequality, and the second graph also suggests that decreases in import tariffs exerts a negative and unequal distributional effect on wages. Steep slopes show high responsiveness of wage inequality with changes in these measures of trade. One of the reasons for the decrease in relative wages of unskilled labour, as tariffs fall, can be that the heavily protected sectors in many developing countries tend to be industries that employ a high proportion of unskilled workers. (Goldberg and Pavcnik 2004)
3 Results
Before presenting second stage results, it is important to test the instruments. Figure 2 shows good results for the relevance of instruments. Nevertheless, usual first order and second order relevance and validity tests are required to have confidence on IV relationships. We have utilized a rich set of openness and trade policy proxies though. Generally for most of the definitions, FR bilateral trade share has not been used before as an instrument. Table 3 presents first stage and higher order tests for the quality of instruments. The 2nd, 3rd and 4th columns present first stage results for all IV regression equations. FR trade shares significantly and positively determine all trade proxies. Though distance from the equator has been insignificant, we do know that location does matter in trade nevertheless. F-Statistics mostly pass Cragg and Donald rule of thumb test. In most cases for trade policy variables, the F-statistics is close to 10. Higher order asymptotic test results are presented in columns 5, 6 and 7. Heteroskedasticity robust estimates are used in the second stage. 2SLS bias is close to 0 in most cases confirming the observation made in figure 2 regarding the suitability and power of IV over a simple OLS. The regressors are exogenous to the error term in all cases as all regression equations reject presence of endogeniety as can be seen from insignificant Sargan Statistics in column 5. Column 9 provides R-square for each corresponding second stage regression equation. High R squares obtained in the second stage with no presence of endogeneity give further credence to the second stage results.
Table 4 provides results for OLS and IV. Column 2 covers all openness and trade
policy variables for 34 regression equations. Column 5 gives results for initial skills representing average years for schooling for 1965. Such early levels of schooling are rightly assumed to be uncorrelated with post 1980 trade reforms and would exogenously determine wage inequality. The results show that initial skill
endowments 65skills relates negatively to inequality suggesting that developing
13
countries that are more educated to begin with do well on wage inequality. This is expected, and in-line with the theory that the countries where skill endowment is more evenly distributed, they are less prone to an adverse distributional effect in relative wages (Fisher 2001; Tuelings and van Rens 2002; Eiche and Cecilia 2001; Bourguignon and Morrisson 1990; Tilak 1989) Not only average years of schooling, but secondary years of schooling for all initial years 1960, 1965, 1970 and 1975 have been found to be negatively related with wage inequality. In contrast higher years of schooling for such initial years are weakly related with wage inequality as found by Mamoon and Murshed (2008)
OLS results depict the same trend of insignificance seen in figure 2 for both
openness and trade policy variables. How ever, by using FR trade shares (Lfrkrom) and distance from the equator (Disteq) as instruments improve the results significantly. The improvement has been observed for all regression equations. All openness and trade policy variables carry expected signs, and nearly all of them bear significant relationship to wage inequality. It would therefore, appear that trade liberalization does worsen the distribution of wages between skilled and unskilled labour in developing countries. If total trade as a percentage of GDP doubles this would lead to a 40 percent increase in wage inequality which is quite significant. However the other variables that capture the importing side of trade have much smaller impact on wage inequality
Above, we have checked the effects of relative factor supply for initial periods when it has been exogenous to trade reforms. Now we come to post reform effect known as skill bias technical change. Trade puts an upward pressure on skill demands. The dynamic trade models suggest that this is also true for developing countries. Many developing countries now invest in education and many have transformed a part of their populations into skilled labour to exploit increased opportunities of good jobs (skilled labour demand) which integration has brought to their door steps. Countries like China and India are good examples. In absolute terms they have more educated people than countries in North East Asia, where education levels are equally distributed among population. However we also know that populations in China and India have been less educated to start with. Education (years of schooling) has been heterogeneous, as only a segment of population (urban) have been provided with increased educational opportunities. Whereas, the uneducated also take part in improved opportunities which international trade provides through industrial sector growth. Most of them form the unskilled of all such sectors. In other words, supply of unskilled labour is far more elastic than supply of skilled labour in both countries. Nowadays, many developing countries have been trying to jump into the Chinese and Indian bandwagon by investing in education. Since developing countries face resource constraint; they can only focus on higher levels of education to exploit the dividends of growth through processes like trade. We would show in
14
coming lines that this phenomenon has inequalities partly because the focus has not been on ‘education for all’.
Table 3 2nd
Order Relevance and Exogeneity Tests for Instruments Variables
Instruments (1st Stage results)
Exogeniety (2nd Stage
Higher Order)
Relevance
(2nd Stage Higher Order)
Endogenous Independent
Lfrkrom
Distq
F
Sargan statistic
Maximal 2SLS Bias (b)
1st Stage heteroskedasticity-robust
N
R2
Openness Variables (Exports +Imports)/GDP, 1985 0.689*** -0.013 40.68*** 0.896 0.000 robust 63 0.67 Import Penetration: overall, 1985 21.073*** -0.186 14.05*** 0.941 0.000 robust 51 0.47 Import Penetration: manufacturing, 1985 13.391*** -0.125 14.82*** 0.925 0.000 robust 51 0.48 Import Penetration: agriculture, 1985 3.544*** -0.010 14.37*** 0.900 0.000 robust 51 0.47 Import Penetration: resources, 1985 3.652*** -0.053 4.05** 0.825 0.003 robust 51 0.20 Import Penetration: overall, 1982 26.717*** -0.343 13.83*** 0.718 0.000 robust 51 0.47 Import Penetration: manufacturing, 1982 13.823*** -0.213 14.88*** 0.671 0.000 robust 51 0.48 Import Penetration: agriculture, 1982 3.792*** -0.013 13.60*** 0.844 0.000 robust 51 0.47 Import Penetration: resources, 1982 8.888*** -0.096 6.53*** 0.759 0.000 robust 51 0.29 TARS trade penetration: overall, 1985 33.041*** -0.599 11.48*** 0.702 0.000 robust 51 0.42 TARS trade penetration: manufacturing , 1985 18.502*** -0.202 10.27*** 0.894 0.000 robust 51 0.39 TARS trade penetration: agriculture, 1985 3.810*** -0.267*** 10.18*** 0.163 0.000 robust 51 0.39 TARS trade penetration: resources, 1985 9.991*** -0.101 4.44*** 0.891 0.001 robust 51 0.22 TARS trade penetration,: overall, 1982 39.081*** -0.773 10.14*** 0.580 0.000 robust 50 0.41 TARS trade penetration: manufacturing , 1982 19.468*** -0.312 10.78*** 0.683 0.000 robust 50 0.41 TARS trade penetration: agriculture, 1982 4.900*** -0.297*** 11.54*** 0.193 0.000 robust 50 0.43 TARS trade penetration: resources, 1982 13.874*** -0.131 4.14** 0.800 0.001 robust 50 0.21 Trade Policy Variables Tariffs Import duties as % imports,1985 -2.565* 0.146 2.27* 0.129 0.083 robust 48 0.13 Tariffs on international inputs and capital goods, 1985 -0.124*** 0.002 4.86*** 0.756 0.001 robust 51 0.24 Trade taxes/ trade, 1982 -0.013 0.0002 0.70 0.914 0.297 robust 27 0.08 Weighted average of total import charges: overall, 1985 -18.861*** 0.344 8.04*** 0.711 0.000 robust 49 0.35 Weighted average of total import charges: manufacturing, 1985
-20.300*** 0.401 8.38*** 0.677 0.000 robust 49 0.36
Weighted average of total import charges: agriculture, 1985
-12.121* 0.164 3.73** 0.867 0.125 robust 49 0.20
Weighted average of total import charges: agriculture, 1985
-26.180*** 1.075** 6.19*** 0.338 0.000 robust 49 0.29
Non Tariff Barriers Non trade barriers frequency on intermediate inputs, 1985
-0.042 0.003 0.68 0.239 0.425 robust 51 0.04
Non-tariff barriers Coverage: overall, 1985 -17.042*** 0.423 4.79*** 0.6133 0.006 robust 49 0.24 Non-tariff barriers Coverage: manufacturing, 1985 -16.367** 0.392 4.29*** 0.643 0.015 robust 49 0.22 Non-tariff barriers Coverage: agriculture, 1985 -12.121* 0.164 3.73** 0.867 0.125 robust 49 0.20 Non-tariff barriers Coverage: resources, 1985 -26.18*** 1.075 6.19*** 0.338 0.000 robust 49 0.29 Composite Measures Sachs and Warner 1980s 0.146 0.003 0.90 0.455 0.266 robust 31 0.09 Measures based on residuals Leamers Measure, 1982 0.162*** -0.0002 5.03*** 0.544 0.001 robust 22 0.46 Gravity-residuals, basic model, 1982 -11.686*** 0.179** 28.01*** 0.436 0.000 robust 36 0.72 Gravity-residuals, augmented model, 1982 -10.863*** 0.141 26.20*** 0.419 0.000 robust 36 0.11 Price Based Measures Black market premium, 1985 0.260 0.012 1.13 0.312 0.187 robust 39 0.08
***, **, * denotes significance at 1%, 5 % and 10% levels respectively
15
Table 4
OLS and IV Regression Results for Theil99Developing
Dependent Variable: Theil99dev
Endogenous Independent Independent
Eq. Openness Variables
OLS IV Initial Skills
OLS IV
1 Lcopen 0.031** (0.014) 0.391** (0.017) Skills65 -0.014**(0.0064) -0.015** (0.006) 2 Impnov85 0.005 (0.004) 0.0016** (0.0007) Skills65 -0.072 (0.0084) -0.015** (0.008) 3 Impnov85m 0.001 (0.0007) 0.0025**(0.001) Skills65 -0.013* (0.008) -0.016** (0.008) 4 Impnov85a 0.0027 (0.003) 0.0095** (0.004) Skills65 -0.011 (0.007) -0.012 (0.008) 5 Impnov85r 0.0002 (0.0017) 0.009* (0.0049) Skills65 -0.011 (0.008) -0.017* (0.010) 6 Impnov82 0.0005 (0.0004) 0.0012** (0.0005) Skills65 -0.012 (0.0078) -0.014* (0.0079) 7 Impnov82m 0.0004 (0.0007) 0.0024** (0.0011) Skills65 -0.011 (0.0079) -0.015* (0.0084) 8 Impnov82a 0.0004 (0.0027) 0.0091**(0.0043) Skills65 -0.011 (0.0079) -0.013 (0.0084) 9 Impnov82r 0.0021** (0.0008) 0.0038** (0.0016) Skills65 -0.012 (0.0074) -0.013* (0.0075) 10 Tars85 0.0005* (0.0002) 0.001** (0.0004) Skills65 -0.013* (0.008) -0.016** (0.008) 11 Tars85m 0.0002 (0.0004) 0.002** (0.001) Skills65 -0.012 (0.0082) -0.019** (0.009) 12 Tars85a 0.0001 (0.0017) 0.005 (0.0028) Skills65 -0.011 (0.008) -0.013 (0.008) 13 Tars85r 0.003*** (0.001) 0.003** (0.0013) Skills65 -0.011 (0.007) -0.011* (0.007) 14 Tars82 0.0003 (0.0002) 0.001** (0.0004) Skills65 -0.012 (0.008) -0.015* (0.008) 15 Tars82m 0.0003 (0.0004) 0.002** (0.001) Skills65 -0.011 (0.008) -0.017* (0.009) 16 Tars82a 0.001 (0.0014) 0.004 (0.003) Skills65 -0.009 (0.008) -0.014 (0.009) 17 Tars82r 0.002*** (0.002) 0.002** (0.001) Skills65 -0.012 (0.007) -0.012* (0.007)
Trade Policy Variables Tariffs 18 Tariffs -0.001 (0.0013) -0.007 (0.005) Skills65 -0.018 ** (0.009) -0.025** (0.012) 19 Owti -0.054 (0.059) -0.255* (0.132) Skills65 -0.011 (0.008) -0.013 (0.009) 20 Txtrdg 0.110 (0.493) 3.825 (2.971) Skills65 -0.009 (0.015) -0.005 (0.026) 21 Totimpv85 -0.001 (0.0004) -0.002** (0.001) Skills65 -0.009 (0.008) -0.011 (0.008) 22 Totimpv85m -0.001 (00004) -0.002** (0.001) Skills65 -0.010(0.008) -0.012 (0.009) 23 Totimpv85a -0.001* (0.001) -0.002** (0.001) Skills65 -0.010 (0.008) -0.011 (0.008) 24 Totimpv85r -0.0003 (0.001) -0.003* (0.001) Skills65 -0.010 (0.008) -0.005 (0.010)
Non Tariff Barriers 25 Owqi -0.038 (0.039) -0.297 (0.289) Skills65 -0.011 (0.001) -0.016 (0.012) 26 Nontar85 -0.0003 (0.0004) -0.002* (0.001) Skills65 -0.011 (0.009) -0.021* (0.012) 27 Nontar85m -0.0003 (0.0003) -0.002* (0.001) Skills65 -0.010 (0.008) -0.023* (0.013) 28 Nontar85a -0.0003 (0.0003) -0.003 (0.002) Skills65 -0.012 (0.008) -0.033* (0.019) 29 Nontar85r -0.0003 (0.0003) -0.001* (0.001) Skills65 -0.010 (0.008) -0.010 (0.008)
Composite Measures 30 Open80s -0.043 (0.035) 0.127 (0.161) Skills65 -0.002 (0.008) -0.010 (0.015)
Measures based on residuals 31 Leamer82 -0.064 (0.063) -0.058 (0.222) Skills65 -0.005 (0.010) -0.010 (0.011) 32 Grmb -0.001 (0.001) -0.001 (0.001) Skills65 -0.010* (0.005) -0.011** (0.005) 33 Grma -0.001 (0.001) -0.001 (0.001) Skills65 -0.010* (0.005) -0.010** (0.005)
Price Based Measures 34 Black 0.023* (0.014) 0.024 (0.048) Skills65 -0.011 (0.007) -0.011 (0.007)
***, **, * denotes significance at 1%, 5 % and 10% levels respectively .Standard errors are presented in parenthesis.
16
ARG
BENBGD
BHR
BOL
BRA
BRB
BWA
CAF
CHL
CHN
CMR
COGCOL
CRI
CYP
DOM
DZA
ECU
EGY
FJI
GHA
GMB
GTM
HKG
HND
HTI
IDNIND IRN
JAM
JOR
KEN
KOR
KWT
LKA
LSO
MEX
MMR
MOZ
MUS
MWI
MYS
NIC
NPL
PAK
PAN
PERPHL
PNG
PRY
RWASDNSEN
SGP
SLV
SWZSYR
TGO
THA
TTO
TUN
TUR
TWN
UGA
URY
VEN
ZAF
ZWE
24
68
10
skilled labor (s
kill
s99)
2 3 4 5 6lcopen(predicted)
Figure 3
Skill Bias Technical Changes
Increased investments in higher education captures the demand side of skill bias
technical change caused by international trade where demand of skills have been rising faster then the supply? Figure 3 shows that trade liberalization augments skills in developing countries. This is true because international trade between developed and developing countries is followed by technology transfer (processes like learning by doing) which improves the general skill level. This means that part of the human capital stock is endogenous to the processes of trade, as hinted at by many endogenous growth models.
A skill bias technical change would also capture the social inequalities in the
populations as we have been arguing that education in most developing countries is heterogeneously distributed among populations. Most are still uneducated. Here a change in skilled human capital, which is endogenous to integration, will have its own independent effect on relative wages. Because trade utilize skilled and unskilled labour but offer higher returns to the skilled relative to unskilled, the effect of education which can be explained through trade should be positive on wage inequality. In other words, this positive effect will be different to that which is attributable to the initial level of human capital endowment.
17
ARGBENBGD
BHR
BOLBRABRBBWA
CAF
CHL
CHN
CMR
COG
COLCRI CYP
DOM
DZA
ECUEGY
FJI
GHA
GMBGTM
HKG
HND
HTI IDN
IND
IRN
JAM
JORKEN
KOR
KWT
LKA
LSO
MEX
MMR
MOZ
MUS
MWI
MYSNICNPL
PAKPANPER
PHL
PNG
PRY
RWASDNSEN
SGPSLV
SWZ
SYR
TGO
THA
TTO
TUN
TUR
TWN
UGA
URYVEN ZAFZWE
0.2
.4.6
.8
Wage inequality(T
heil index)
2 4 6 8 10skilled labor (skills99)
AGO
ARG
BDI
BENBFA
BGD
BHR
BHS BLZBOLBRA
BRBBTN
BWA
CAF
CHL
CHN
CIV
CMR
COG
COLCPV
CRICUB
CYP
DOM
DZA
ECUEGY
ETHFJI
GABGHA
GMBGNQGTM
HKG
HND
HTIIDN
IND
IRN
JAM
JORKEN
KOR
KWT
LKA
LSO
MAR
MDGMEX
MMR
MNG
MOZ MRT
MUS
MWI
MYS NAM
NGA
NICNPL
OMNPAK
PANPERPHL
PNGPRI
PRY
QAT
RWA
SAU
SDNSEN
SGPSLV
SUR
SWZ
SYC
SYR
TGO
THA
TTO
TUN
TUR
TWN
TZA
UGA
URY
VCT
VEN
YEM
ZAF ZWE
0.2
.4.6
.8
Wage inequality(T
heil index)
4.6 4.8 5 5.2 5.4 5.6skilled labor (predicted skills99)
Figure 4
Skills versus Skill Bias Technical Change
Figure 4 shows two graphs. The first one illustrates a simple relationship between skill levels in 1999 and wage inequality, and suggests that countries with higher stocks of skilled labour will have less inequality. The second graph has a different measure of skills (based on predicted values by regressing skill99 on FR trade shares; or those skills which can be measured in relation with trade), conversely suggests that skill accumulation due to greater global integration raises wage inequality. It implies that the skills that accrue directly through the processes of trade, contribute to wage inequality. This is in line with Tinbergen’s (1975) arguments, and the earlier discussion regarding the dual role of skilled human capital in the economy. In order to examine the dual (positive and negative) impact of skilled labour stocks on the skilled-unskilled labor wage differential, it is necessary to modify the basic ‘wage inequality model’ by introducing an interaction term between skills and some selected openness and trade policy measures.
Conceptually speaking, the present wage inequality framework should contain:
)](),([( 0skillsSkillsopennessSkillsfInequality ii= (9)
(+) (-)
)](),([( 0skillsSkillsytradepolicSkillsfInequality ii= (10)
(+) (-)
18
Skills which depend on Initial skill endowments have a negative effect on wage inequality in Eqs (9) and (10). The skills that accrue through trade liberalization exacerbate inequality.
The econometric form of the modified wage inequality model is:
iiiiSkillsSkillsOPENnInteractioTHEIL 51113 99)99( ερλ ++×Ω+= (11)
iiiiSkillsSkillsTPnInteractioTHEIL 73335 99)99( ερλ ++×Ω+= (12)
99Skills represents skill endowments in 1999; a point in time when earlier trade
liberalization had had time to play a role in skill accumulation in developing countries
through technology transfer. Hence 99skills would depend on initial skill levels and
trade. To isolate the effect of trade from initial skill endowment effect in 99skills , we
need to first find instruments which can separately explain 99skills in terms of trade
and initial skill levels. Mamoon (2008) finds that drop out rates (Drop) and schooling
days in a year (Schday) exogenously determine average years of schooling for 1999
(Skills99). Here, we can instrument Skills99 with Drop80 and Schday. Such exercise
will enable us to explain Skill99 only with reference to initial endowments. To
capture the effect of trade on Skill99, we first interact Skills99 with some selected
proxies of openness and trade policy to form Interaction (OPENxSkills99) and
Interaction (TPxSkills99). Following the observation made in figure 4 and to ensure
that only trade can explain this interaction term, we instrument it on FR (1999)
bilateral trade shares (Lfrkrom) and distance from the equator. The interaction terms
capture the effect of skills on inequality, while taking into account the extent to which
each developing country has integrated with world markets. In the first stage
Interaction (OPENxSkills99), Interaction (TPxSkills99) and Skills99 would be
regressed over all four instruments identified above.
=× iSkillOPENnInteractio 2)99(
iiiii DisteqSchdayDropFR 911111 80 εγκθ +++++∂ (13)
=× iSkillTPnInteractio 2)99(
iiiii DisteqSchdayDropFR 1022222 80 εγκθ +++++∂ (14)
iiiiii DisteqSchdayDropFRSkill 11333332 8099 εγκθ +++++∂= (15)
Here iFR stands for predicted trade shares from gravity equations computed by
Frankel and Romer (1999), Drop80 stands for dropout rates in 1980, Schday
represents schooldays in a year whereasiDisteq (distance from the equator) is a proxy
for geography.
There is evidence that suggests unequal investments in higher education are
important determinants of increasing inequalities in developing countries (Barro
19
1999). We want to know whether it is overall low education levels which are driving
wage inequality in developing countries or it is that within education sector, it is
higher level of education which is more closely related with skill bias technical
change. According to Acemoglu (1999), trade induced skill bias technical change
causes higher wage inequalities in developing countries through a wage premium
effect induced towards higher levels of education. (Acemoglu, 2001)
Here we are not only to test for effect of international trade on average levels of
skills (skill99) but also on higher levels of skills (Hyr99). This provides us with
additional interaction terms based on later category of skill levels. The latter category
offers more-specific information about the relationship between inequality and higher
levels of education. Equations (13), (14) and (15) would remain the same in the right
hand side for higher skills (Hyr99). Only change in the left hand side would be that
Hyr99 would be used instead of Skills99. The same change is due in equations (11)
and (12).
Table 5 1st Stage Regression results for Instrumental variables
First Stage Results: Interactions Between Openness/ Trade Policy and average Skills
Lcopen × Skill99
Impen85 × Skill99
Impen82 × Skill99
Tars85 × Skill99
Tars82 × Skill99
Tariffs × Skill99
Owti × Skill99
Totimpv85 × Skill99
Owqi × Skill99
Nontar85× Skill99
Skill90 Lfrkrom 3.718 110.70 139.72 164.63 196.251 -16.769 -0.547 -75.181 -0.243 -70.47 0.0144 (2.43)** (3.82)**
* (3.67)***
(3.18)***
(2.96)***
(-1.86)* (-3.2)*** (-3.6)*** (-0.94) (-2.7)*** (0.04)
Drop80 -0.203 -2.918 -3.356 -5.865 -6.242 -0.115 -0.004 -0.0063 -0.0037 -0.796 -0.039 (-3.7)*** (-2.7)*** (-2.54)** (-3.8)*** (-2.7)*** (-0.32) (-0.65) (-0.01) (-0.39) (-0.69) (-3.2)*** Schday -0.029 1.26 1.208 0.947 1.152 0.5491 -0.001 -0.247 0.0005 0.464 -0.006 (-0.45) (1.00) (0.72) (0.42) (0.39) (1.10) (-0.18) (-0.27) (0.05) (0.36) (-0.45) Disteq 0.063 0.063 -1.161 -1.875 -3.391 1.164 0.009 1.603 0.017 1.853 0.034 (0.61) (0.03) (-0.46) (-0.54) (-0.77) (2.02)** (0.84) (1.20) (1.04) (0.98) (1.50) N 64 55 54 55 53 47 54 52 54 52 64 F-statistics 6.42*** 7.35*** 7.18*** 6.36*** 5.72*** 2.30* 3.14** 4.34*** 0.68 2.60*** 4.15***
R-square 0.30 0.37 0.37 0.34 0.32 0.18 0.21 0.27 0.05 0.18 0.219
First Stage Results: Interactions Between Openness/ Trade Policy and Higher Skills
Lcopen × Hyr99
Impen85 × Hyr99
Impen82 × Hyr99
Tars85 × Hyr99
Tars82 × Hyr99
Tariffs × Hyr99
Owti × Hyr99
Totimpv85 ×Hyr99
Owqi ×Hyr99
Nontar85× Hyr99
Hyr99 Lfrkrom 0.1859 6.635 7.940 9.539 10.976 -16.04 -0.026 -3.87 -0.007 -3.210 0.00001 (1.22) (4.00)**
* (3.82)***
(13.2)***
(3.12)***
(-1.78)* (-2.32)** (-2.33)** (-0.38) (-1.52) (0.00)
Drop80 -0.0061 -0.109 -0.115 -.229 -0.229 -0.085 -0.0003 0.032 -0.0001 0.036 -0.0009 (-1.12) (-1.79)* (-1.60) (-2.11)** (-1.88)* (-0.24) (-0.08) (0.50) (-0.16) (0.45) (-0.73) Schday -0.001 0.135 0.174 0.121 0.204 0.746 -0.0002 -0.01 -0.0001 0.006 -0.0005 (-0.11) (1.82)* (1.85)* (0.92) (1.28) (1.42) (-0.35) (-0.14) (-0.14) (0.06) (-0.32) Disteq 0.012 0.138 0.145 0.137 0.151 1.072 0.001 0.103 0.0005 0.152 0.004 (1.17) (1.24) (1.05) (0.67) (0.64) (1.85) (0.91) (0.94) (0.41) (1.10) (1.47) N 62 54 53 54 52 46 53 51 53 51 62 F-statistics 1.31 7.09*** 7.00*** 5.06*** 5.12*** 2.28* 1.87 2.00 0.11 1.14 0.87 R-square 0.08 0.37 0.37 0.29 0.30 0.18 0.13 0.15 0.009 0.09 0.05
t- Values in the parenthesis. ***, **, * denotes significance at 1%, 5 % and 10% levels respectively
20
Table.6
Higher Order Relevance and Exogeniety Tests for Theil99dev
Relevance Exogeneity
Eq.
Endogenous Dependent
Variable:
Wage Inequality in Developing Countries (Theil99dev)
N 1st Stage
heterosk
edasticit
y-robust
Maxim
al
2SLS
Bias
(b)
Cragg-
Donald
N*minE
val stat.
Chi-sq()
Anderson-
Rubin test
of joint
significanc
e of
endogen-
ous
regressors
F-Statistic
Sargan statistic
(overidentificati
on test of all
instruments)
Chi-Sq()
Endogenous Independent Variables:Interactions Between Openness/ Trade Policy and Average Skills for 1999
(Instruments= Disteq, Lfrkrom, Drop80, Schday)
35 Interaction (Lcopen × Skill99) 62 Robust 0.0005 17.76** 1.91 2.400
(0.301)
36 Interaction (Impen85 × Skill99) 55 Robust 0.0028 14.11** 2.29* 4.390
(0.111)
37 Interaction (Impen82 × Skill99) 54 Robust 0.0037 13.46** 1.88 3.316
(0.191)
38 Interaction (Tars85 × Skill99) 55 Robust 0.0027 14.16** 2.29* 3.795
(0.149)
39 Interaction (Tars82 × Skill99) 54 Robust 0.1699 5.03** 2.28* 5.265
(0.072)*
40 Interaction (Tariffs × Skill99) 47 Robust 0.2078 4.55** 2.61** 1.105
(0.574)
41 Interaction (Owti × Skill99) 54 Robust 0.0037 13.51** 2.27* 1.678
(0.432)
42 Interaction (Totimpv85 ×
Skill99)
52 Robust 0.007 11.86** 2.37* 2.04
(0.360)
43 Interaction (Owqi × Skill99) 54 Robust 0.5023 2.35** 2.27* 0.628
(0.730)
44 Interaction (Nontar85× Skill99) 52 Robust 0.0145 10.55** 2.37* 0.717
(0.698)
Endogenous Independent Variables:Interactions Between Openness/ Trade Policy and Higher Skills for 1999 (Instruments= Disteq, Lfrkrom, Drop80, Schday)
45 Interaction (Lcopen × Hyr99) 62 Robust 0.286 3.78** 1.83 3.06
(0.215)
46 Interaction (Impen85 ×Hyr99) 54 Robust 0.167 5.06** 2.28* 6.295
(0.043)**
47 Interaction (Impen82 × Hyr99) 53 Robust 0.1279 5.69** 1.88 5.132
(0.076)*
48 Interaction (Tars85 × Hyr99) 53 Robust 0.0033 13.75** 1.83* 3.096
(0.212)
49 Interaction (Tars82 × Hyr99) 52 Robust 0.085 6.62** 1.84* 4.439
(0.110)
50 Interaction (Tariffs × Hyr99) 46 Robust 0.933 0.43 2.72** 0.040
(0.980)
51 Interaction (Owti × Hyr99) 53 Robust 0.082 8.25** 2.25* 3.779
(0.286)
52 Interaction (Totimpv85 ×
Hyr99)
51 Robust 0.395 2.98** 2.34* 2.350
(0.308)
53 Interaction (Owqi × Hyr99) 53 Robust 0.955 0.33 2.25* 0.204
(0.903)
54 Interaction (Nontar85× Hyr99) 51 Robust 0.348 3.29** 2.34* 1.059
(0.5669)
t- Values in the parenthesis. ***, **, * denotes significance at 1%, 5 % and 10% levels respectively
21
To investigate integration, six definitions of openness: Lcopen, Impnov85,
Impnov82, Tars85 and Tars82 and five different proxies of trade policy namely
Tariffs, Owti, Totimpov85, Owqi, and Nontarr87 are used (see summary statistics
provided in table 2 for details about the variables).
Table 5 provides results for the 1st stage. FR bilateral trade shares (Lfrkrom) and
(Drop80) significantly explain the endogenous regressors. F-statistic is less than 10 in
all cases; violating Cragg-Donald rule of thumb. However, we are dealing with 4
instruments and thus a first order tests are insufficient and higher order asymptotic
tests are required. Table 6 shows that higher order Crag Donald F-Statistics, which
also corresponds to 2SLS bias, do relatively better in case of Skills99 than in case of
Hyr99. Largely for Skills99, 2SLS bias has been 0 or close to it. Only for equation 43,
higher order asymptotes show a large 2SLS bias in favour of OLS. For Hyr99, 2SLS
bias is large enough to fail the test for at least 4 equations (50, 52, 53 and 54). Over
identification problem is not present for most of the equations for the interaction
terms of Skills99 or Hyr99. Good results for higher order asymptotic tests establish
the statistical validity of the model while heteroskedasticity robust IV estimates are
obtained for the second stage. The omitted variable bias has thus been addressed in all
regression equations.
Table 7 presents results for Interaction (OPEN×Skills99), Interaction
(TP×Skills99) and Skills99. Interaction terms always enter wage inequality
significantly at 5% and 10% levels of significance. Interaction(OPEN× Skills99) is
positively related with wage inequality whereas Interaction (TP×Skills99) is negatively related with wage inequality confirming the presence of skill bias technical change for developing countries where skill demand has resulted in payment of skill premiums causing wages of skilled (educated) to rise at greater rates than the wages of unskilled (uneducated). 1st column shows that Skills99 are negatively related with wage inequality. Remember Skills99 have been instrumented by Drop80 and Schday, and thus represents post reform skills which only depend on the initial skills that have been prevailing in developing countries. Results confirm that countries that have integrated to world markets with skills (levels of education) that were homogenously distributed among the population have witnessed a fall in wage inequality. However for other measures of openness and trade policy, Skills99 have been insignificant. That may be due to the reason that other trade proxies represent the importing sector. For most developing countries, importing sector is more closely related with the manufacturing sector and the technical bias effect of trade would be more prominent in importing sectors. Importing sectors employ more skilled labour. For such cases interaction terms are over riding explanatory power of results with respect to Skills99.
22
Table 7 Interactions between Openness/ Trade Policy and Average Skills
Independent
Variables
Dependent Variable: Theil99dev
1 2 3 4 5 6 7 8 9 10
Interaction Terms
Lcopen × Skill99 0.0062
(2.25)**
Impen85 × Skill99 0.0002
(1.67)*
Impen82 × Skill99 0.00013
(1.64)*
Tars85 × Skill99 0.00012
(1.87)*
Tars82 × Skill99 0.0001
(1.64)*
Tariffs × Skill99 -0.002
(-1.82)*
Owti × Skill99 -0.053
(-1.9)**
Totimpv85 × Skill99 -0.001
(-2.1)**
Owqi × Skill99 -0.087
(-1.30)
Nontar85× Skill99 -0.002
(1.96)**
Average Skills, 1999
Skills99 -0.0279 -0.0034 -0.001 -0.005 -0.002 0.031 0.013 0.011 0.021 0.018
(-1.80)* (-0.27) (-0.05) (-0.41) (-0.17) (1.09) (1.24) (0.95) (0.16) (1.25)
N 64 55 54 55 53 47 54 52 54 52
F 2.42* 1.46 1.45 1.80* 1.42 1.56 2.14* 2.48** 0.98 2.15*
R2 0.63 0.51 0.53 0.52 0.52 0.18 0.42 0.48 0.32 0.28
t- Values in the parentheses.***, **, * denotes significance at 1%, 5 % and 10% levels respectively Standard errors are corrected for as run Durbin-Wu-Hausman test (augmented regression test) for endogeneity (see Davidson
and MacKinnon. 1993)×
Does wage inequality in developing countries also follow the ‘Pareto principle’
where a rise in wages of ‘higher educated’ would be more than ‘only educated’? Are we capturing an element of within group inequality also when we say that wage inequality between skilled and unskilled is rising? Because a rise in within group inequality can also capture a rise in wages of skilled labour which would be disproportional to not only less educated (low skilled) but also uneducated (unskilled). Is technical bias affecting the wages of highly skilled or “the proportional rise in returns to skills is across the board (for all education levels)”? By answering these questions, we can know ‘what level of education matters most’.
23
Table 8
Interactions between Openness/ Trade Policy and Higher Skills
Independent
Variables
Dependent Variable: Theil99dev
1 2 3 4 5 6 7 8 9 10
Interaction Terms
Lcopen × Hyr99 0.1001
(1.85)*
Impen85 ×Hyr99 0.002
(1.34)
Impen82 × Hyr99 0.002
(1.35)
Tars85 × Hyr99 0.002
(1.67)*
Tars82 × Hyr99 0.002
(1.49)
Tariffs × Hyr99 -0.004
(-0.82)
Owti × Hyr99 -1.013
(-1.9)*
Totimpv85 × Hyr99 -0.007
(-2.2)**
Owqi × Hyr99 -3.221
(-0.59)
Nontar85× Hyr99 -0.01
(-1.89)*
Higher Skills, 1999
Hyr99 -0.408 -0.056 -0.093 -0.067 -0.106 1.066 0.219 0.123 0.4601 0.213
(-1.53) (-0.33) (-0.63) (-0.40) (-0.73) (0.58) (1.17) (0.60) (0.56) (0.81)
N 62 54 53 54 52 46 53 51 53 51
F 2.63* 0.85 0.98 1.31 1.14 0.37 2.06* 2.39* 0.19 2.75*
R2 0.61 0.54 0.58 0.54 0.56 - 0.46 0.48 - 0.21
t- Values in the parentheses. ***, **, * denotes significance at 1%, 5 % and 10% levels respectively Standard errors are corrected for as run Durbin-Wu-Hausman test (augmented regression test) for endogeneity (see Davidson and MacKinnon. 1993)
We already find evidence that by ignoring ‘education for all’, developing countries
have increased the wage gap among the labour force based on skill differentials. Here we examine whether skill biased technical change benefits higher levels of education
more. Interaction(OPEN×Hyr99) and Interaction(TP×Hyr99) will capture the trade effect of higher education on rising skill returns in developing countries. Table 8 finds supporting evidence that higher skills are related with wage inequality in developing countries. Higher coefficients for all cases when significance is achieved (columns 1, 4, 7, 8 and 10) for ‘Hyr99 interaction terms’ in comparison to coefficients for ‘Skills99 interaction terms in table 7’ suggest that wage inequality is more sensitive towards higher levels of education.
24
For example we know from table 4 that with each additional year of initial levels of schooling in a country, wage inequality is expected to fall by 10 percent. In table 7, with each additional year of schooling post liberalization, wage inequality is expected to rise by less than 1 percent. However with each additional year of higher levels of schooling (post secondary) post trade liberalization can double wage inequality. This shows that wage inequality is more sensitive to higher levels of education than all levels of education. This result may also confirm then that preference to higher education in developing countries is causing distortions in labour markets where following “Pareto Principle” the labour with higher levels of education earn more than just educated while uneducated are worse off.
In this paper, our empirical frame work has actually captured the disproportional
rise in returns to education (rising wage inequality) due to trade liberalization which favours education and higher levels of education even more. This is to say that we also find that returns to schooling rise with additional years of schooling and the relationship is a dynamic one to determine rising wage inequality in developing countries. Higher levels of schooling are more sensitive to wage inequality than over all levels of education.
The non linear relationship between different levels of education and its economic
outcomes can also be extended to social returns to education and its potentially dual role. The different effects of different levels of education on economic development in this context are quite nicely stated by Krueger and Lindhal (2001): ‘ The social return (to education) can be higher because of externalities from education, which could occur, for example, if higher education leads to technological progress that is not captured by in the private return to that education, or if more education produces positive externalities, such as a reduction in crime and welfare participation or more informed political decisions. The former is more like if human capital (education) is expanded at higher levels of education while the latter is more likely if human capital (education) is expanded at lower levels’ (1107)
Developing countries can invest in higher levels of education to exploit social
externalities which can generate and sustain technical progress amid globalization. However, social returns to education by raising overall education level may carry more deep rooted positive effects in the economy. Our results strongly support in favor of raising over all education levels in the society The education bias of trade liberalization can be exploited in favour of the poor in a country through investments in all levels of education. That is one way to make trade induced growth good for the poor.
The unequal returns to education in a dynamic trade fame work may partly explain
why initial levels of education (captured by endogenously determined average years of schooling) are more closely related with good economic outcomes than higher education attainment (endogenously determined improved level of schooling) in
25
developing countries as also pointed out by other studies (see i.e., Krueger and Lindahl, 2001). This is not to say that basic education in itself as a target is enough. When the economy begins to trade more with the outside world, perhaps there is a need for a balanced education policy in the South. Providing the necessary higher/technical education should not, however compromise the achievement of primary education. Investing in higher education may not be as effective as investing in overall education levels of the population. Primary education is the first step towards a more skilled labour force, and it is a pro-poor policy, as the overwhelming majority of the poor remain uneducated. (Mamoon 2005)
In the immediate time frame, in order to minimize the positive effects of trade on
wage inequality in developing countries, one solution is to trade among each other, where more trade may take place in low skill intensive goods but developing countries can also trade in relatively skill intensive produce. Developing countries lie in a heterogeneous plain where some may be following relatively skilled intensive (in local terms) production activities. Regional trade can absorb the negative effect of skill bias technical change. However in the longer run, investment in education sector would remain the key factor to effect wage dispersion in developing countries. More resources in education sector need to be channeled to secondary and primary education levels in addition to higher education as suggested by Acemoglu (2001).
4 Conclusions
In summary, the discussion here suggests that the earning inequalities in developing countries have two important determinants. First, there are significant entrenched inequalities in educational attainment. Second, increased international trade transforms these education inequalities into wage inequalities by favouring skilled labour over unskilled labour. In line with previous studies, this analysis found that education might be central to explaining the increasing gap in relative wages between skilled and unskilled workers in developing countries. Although the analysis supports the argument that those countries with a higher initial level of human capital do well on the inequality front, it also suggests that human capital, particularly the part related to higher education, which accrues after trade liberalization has in-egalitarian effects. Governments in developing countries tend to invest more in higher education at the cost of primary education in order to seize short-term benefits from globalization.
One reason for this bias towards higher education, in education policies in developing countries may lie in the belief that elementary education has a very limited direct role in determining growth rates. According to Barro (1999), the rate of economic growth responds more to secondary or higher education levels rather than elementary schooling. International trade in manufacturing and services is one of the key engines for growth. This often requires college graduates or those who have at least finished high school. One reason why India and China became havens for international outsourcing and trade is that they have managed to accumulate relatively
26
educated and skilled human capital by investing in higher education.1 In the last two years, the United States has lost two million manufacturing jobs to China, India and other third world countries, mainly in the Far East. Consequently, in 2001, the manufacturing sector in the US shrank to just 18 per cent of GNP from 48 per cent in 1950 and it is expected to recede to just 10 or 11 per cent within the next 10 years2. Forrester Research, a market research firm predicts that at least 3.3 million white-collar jobs and $136 billion in wages will shift from the US to low-cost countries by 2015 and most of them will find their way to Indian or Chinese centers3. Countries in the South that are set to benefit most from globalization are those that transformed at least a segment of their labour force into a relatively skill intensive force by investing in higher education programs. Therefore, it is no surprise that in order to be competitive, many developing countries have a tendency to invest in higher education at the cost of primary education to achieve greater growth. For example, Pakistan’s current education policy skews towards higher education. In fiscal year 2003, the government increased its higher education budget to Rs 5 billion from Rs 800 million five years ago—an increase of nearly 400 per cent. In 2004, the government doubled the previous year’s expenditure on higher education. In contrast, the budget for primary education increased by a meager average of 4 per cent per annum for the last few years. This apparent pro-growth higher education policy of Pakistan at the expense of primary education may very well accelerate growth, but it excludes the poor and unskilled and will likely lead to increased wage and income disparities in the country.
As noted earlier, governments in developing countries tend to focus their education policies on higher education in anticipation that this investment will yield quick dividends in the current international business environment. Although they are right, they need to realize that promoting higher education at the cost of primary education breeds greater income inequality in the absence of countervailing policies and is not pro-poor. Consequently, governments need to increase the mean level of human capital through a balanced education policy where primary education is valued as much as higher education. An equitable education policy will decrease the skilled-unskilled wage premium, as the overall supply of low skilled and uneducated workers decreases and the supply of educated workers increases, as well as give rise to general equilibrium effects that reduce wage inequality. Additionally such policies are consistent with the millennium development goals (MDGs). This researcher agrees with the recent World Development Report (2006), which suggests that wider access to education and jobs could ‘level the economic playing field’ and improve livelihoods.
! ""# ""#"!
27
These results also have implications for the speed at which trade policies are liberalized. Developing countries require time to adjust to an open regime because of their low-skill level stocks; the implication being that better educated nations should liberalize faster. If skill levels were homogenously distributed, free trade would lead to a more-equal and pro-poor growth. In the short-run, it may be wiser for developing economies to first look for regional trade agreements by following the model of the European Union; something that is not easy to achieve amongst poor countries who contemporaneously do not trade much with each other. But there may be potential in that direction. The rationale for regional trade is that developing countries lie in a heterogenic plain where each country is located at different rungs of the technical ladder. To climb the ladder, the more efficient way may lie in increased South-South trade rather than North-South trade in a regional framework as in the case of ASEAN (Association of South East Asian Network). Similar trade agreements would enable developing countries to liberalize their economies slowly, while at the same time enabling their workforce to ascend the skill ladder (Mamoon, 2004)
28
References:
Acemoglu, D. (2001) ‘Human Capital Policies and the Distribution of Income: A Framework
for Analysis and Literature Review,’ Treasury Working Paper Series 01/03 http://ideas.repec.org/p/nzt/nztwps/01-03.html
Acemoglu, D. (1999) ‘Patterns of Skill Premia,’ NBER Working Papers 7018, National
Bureau of Economic Research, Inc
Acemoglu, D., S. Johnson and J. Robinson (2004) ‘Institutions as the Fundamental Cause of
Long-Run Growth,’ NBER Working Papers 10481, National Bureau of Economic Research, Inc.
Alcala, F. and A. Ciccone (2002) ‘Trade and Productivity’, CEPR Discussion Paper No.
3095.
Amann, E., N. Aslnindis, F. Nixon and B. Walters (2002) ‘Economic Growth and Poverty
Alleviation: A Reconsideration of Dollar and Kraay’, Development Economic Discussion Papers, No. 0407, http://les.man.ac.uk/ses/research/devpapers/Povertyconference%5B1%5D.pdf.
Arbache, J.S., A. Dickerson and F. Green (2004) ‘Trade Liberalization and Wages in
Developing Countries’, Economic Journal 114(493): 73-96.
Aghion, P., C. Eve and C. Garcia-Penalosa (1999) ‘Inequality and Economic Growth: The
Perspective of the New Growth Theories’, Journal of Economic Literature XXXVII: 1615-60 (December).
Barro, R. J. (1999) ‘ Does and Income Gap put a Hex on growth,’ Business Week, 29 March
Barro, R. J. and J. W. Lee (2001) ‘International Data on Educational Attainment: Updates and Implications,’Oxford Economic Papers, Oxford University Press, 53 (3), 541-63 Barro, R. J. and J. W. Lee (1996) ‘International Measures of Schooling Years and Schooling Quality,’ American Economic Review, American Economic Association,.86 (2), 218-23 Behrman, J.R., N. Birdsall and M. Szelely (2001) ‘Pobreza, Desigualdad, y Liberalizacion
Comercial y Financiera en America Latina’, Inter American Development Bank (IADB), Washington DC, http://www.undp.org/rblac/liberalization/docs/capitulo3.pdf.
Bourguignon, F. and C. Morrison (1999) ‘Income Distribution, Development and Foreign
Trade: A Cross Section Analysis’, European Economic Review 34(6): 1113-32.
29
Cockburn, J. (2002) ‘Trade Liberalization and Poverty in Nepal: A Computable General
Equilibrium Micro Simulation Analysis’, The Centre for the Study of African Economies Working Paper Series. Working Paper 170, http://www.bepress.com/csae/paper170.
Chen, S. and M. Ravallion (2003) ‘Household Welfare Impacts of China’s Accession to the
World Trade Organization’, World Bank Policy Research Working Paper 3040, www.econ.worldbank.org/files/26013_wps3040.pdf.
Dollar, D. and A. Kraay (2004) ‘Trade, Growth, and Poverty’, Economic Journal 114: F22-
F49.
Dollar, D. and A. Kraay (2002). ‘ Growth Is Good for the Poor,’ Journal of Economic
Growth, Springer, vol. 7(3), pages 195-225,
Edwards, S. (1998 ) ‘Openness, Productivity and Growth: What Do We Really Know?’,
Economic Journal 108: 383-98.
Eiche, T.S. and G.P. Cecilia (2001) ‘Inequality and Growth: The Dual Role of Human Capital
in Development’, Journal of Development Economics 66(1): 173-97.
Frankel, J. and D. Romer (1999) ‘Does Trade Cause Growth?’ American Economic Review
89(3): 379-99.
Fischer, R. D. (2001) ‘The Evolution of Inequality after Trade Liberalization’, Journal of
Development Economics 66: 555-79.
Friedman, J. (2000) ‘Differential Impacts of Trade Liberalisation on Indonesia’s Poor and
Non-Poor’, Conference on International Trade and Poverty, Stockholm (20 October).
Goldberg, Pinelop, K. and Pavcnik, Nina (2004), ‘Trade, Inequality and Poverty: What Do We Know? Evidence From Recent Trade Liberalization Episodes in Developing Countries,’ Conference Paper Brooking Trade Forum, 2004
http://www.econ.yale.edu/~pg87/brookings.pdf
Greenaway D., W. Morgan and P. Wright (2001) ‘Trade Liberalisation and Growth in Developing Countries’, Journal of Development Economics 67: 229-34.
Hanson, G. and A. Harisson (1999) ‘Trade and Wage Inequality in Mexico’, Industrial and
Labor Relations Review 52(2): 271-88.
Harrison, A. (1996) ‘Openness and growth: A Time-Series, Cross Country Analysis for
Developing Countries,’ Journal of Development Economics, 48, 419-447
Hiscox, M.J. and Kastner, S. L. (2002)’A General Measure of Trade Policy Orientations:
Gravity-Model-Based Estimates for 83 Nations, 1960 to 1992.’ UCSD Working Paper
30
Jayasuriya, S. (2002) ‘Globalisation, Equity and Poverty: The South Asian Experience’,
Annual Global Conference of GDN, Cairo, Egypt.
Krueger, A. B. and M. Lindahl (2001) ‘ Education for Growth: Why and For Whom?’
Journal of Economic Literature, (39), 1101-1136
Leamer, E. (1988) ‘Measures of Openness’ in (R. Baldwin, ed.) Trade Policy and Empirical
Analysis, Chicago: University of Chicago Press
Legovini A., C. Bouillon and N. Lustig (2001) ‘Can Education Explain Income Inequality
Changes in Mexico?’, Inter-American Development Bank (IADB), Washington DC.
Lofgren, H. (1999) ‘Trade Reform and the Poor in Morocco: A Rural-Urban General
Equilibrium Analysis of Reduced Protection’, TMD Discussion Paper No. 38, Washington DC: IFPRI.
Mamoon, D. and S.M. Murshed (2008) ‘Unequal skill premiums and trade liberalization: Is
education the missing link?’ Economics Letters, 100(2): 262-266
Mamoon, D. (2008) ‘Trade, Poverty, Inequality and Security’, Phd Dissertation, Shaker
Publishing Mamoon, D. (2005) ‘Education for all is Central to Higher Education Reforms in Developing
Countries’, in A. Boeren and Gerrit Holtland (eds), A Changing Landscape - Making
Support to Higher Education and Research in Developing Countries more Effective, the Netherlands: NUFFIC, http://www. nuffic.net/common.asp?id=1704.
Mamoon, D. (2004) ‘Poverty and Prosperity in Pakistan’, Asia Times, Honk Kong (24
March), http://www.atimes.com/atimes/South_Asia/FC25Df06. html /
Pritchett, L. (1996) ‘Measuring outward orientation in LDCs: Can it be done?’ Journal of
Development Economics, 49 (2): 307-335 Ravallion, M (2003), ‘The Debate on Globalization, Poverty, and Inequality: Why
Measurement Matters’, World Bank Policy Research Working Paper 3038, Washington DC, http://econ.worldbank.org/files/26010_wps3038. pdf.
Rodriguez, F. and D. Rodrik (2000) ‘Trade Policy and Economic Growth: A Skeptic’s Guide
to the Cross-National Evidence’, NBER Macroeconomics Annual 2000, 15: 261.
Rodrik, D., A. Subramanian and F. Trebbi (2004) ‘Institutions Rule: The Primacy of
Institutions Over Geography and Integration in Economic Development’, Journal of
Economic Growth 9(2): 131-65.
31
Rose, A. K. (2004) ‘Do WTO Members Have a More Liberal Trade Policy?’ Journal of
International Economics, 63(2): 209-235.
Sachs, J.D. and A. Warner (1995) ‘Economic Reform and Process of Global Integration’,
Brookings Papers on Economic Activity 1: 1-118.
Tavares, J. (1998) Trade Openness, Factor Proportions and Politics, mimeo, Harvard
University.
Tinbergen, J. (1975) Income Distribution: Analysis and Policies, Amsterdam: American
Elsevier.
Tilak, J.B.G. (1989) Education and its Relation to Economic Growth, Poverty and Income
Distribution, Washington DC: World Bank.
Teulings, C.N. and T.V. Rens (2002) ‘Education, Growth and Income Inequality’, CESifo
Working Paper Series No. 653; Tinbergen Institute Working Paper No. 2002-001/3.
Winters, L.A. (2004) ‘Trade Liberalization and Economic Performance: An Overview’, The
Economic Journal 114: 4-21.
Wood, A. and C. Ridao-Cano (1999) ‘Skill, Trade and International Inequality’, Oxford
Economic Papers 5(1): 89-119.
World Bank(2001) A Globalised Market – Oppertunities and Risk for the Poor, Global Poverty Report, Washington DC World Development Report (2006) Equity and Development, Washington DC: World Bank
32
Table 1 List of Countries for Theil Index (Theil99devloping) Afghanistan (1988) Algeria (1997) Angola (1993) Argentina (1996) Bahamas, The (1990) Bahrain (1992) Bangladesh (1990) Barbados (1997) Belize (1992) Benin (1981) Bhutan (1989) Bolivia (1997) Botswana (1997) Brazil (1994) Burkina Faso (1981) Burundi (1990) Cameroon (1997) Cape Verde (1993) Central African Republic (1993) Chile (1997) China (1985) Colombia (1997) Congo, Rep. (1988) Costa Rica (1997) Cote d'Ivoire (1997) Cuba (1988) Cyprus (1997) Dominican Republic (1985)
Ecuador (1997) Egypt, (1997) El Salvador (1997) Equatorial Guinea (1990) Eritrea (1988) Ethiopia (1997) Fiji (1997) Gabon (1994) Gambia, The (1981) Ghana (1995) Guatemala (1997) Haiti (1988) Honduras (1994) Hong Kong, China (1997) India (1997) Indonesia (1997) Iran, Islamic Rep (1993) Iraq (1985) Jamaica (1990) Jordan (1997) Kenya (1997) Korea, Rep. (1997) Kuwait (1997) Lesotho (1994) Liberia (1985) Libya (1980) Macao, China (1997) Madagascar (1988) Malawi (1997)
Malaysia (1997) Mauritania (1978) Mauritius (1997) Mexico (1997) Moldova (1994) Mongolia (1994) Morocco (1997) Mozambique (1994) Myanmar (1997) Namibia (1994) Nepal (1996) Nicaragua (1985) Nigeria (1994) Oman (1997) Pakistan (1996) Panama (1997) Papua New Guinea (1989) Paraguay (1991) Peru (1994) Philippines (1997) Puerto Rico (1997) Qatar (1994) Rwanda (1985) Saudi Arabia (1989) Senegal (1997) Seychelles (1988) Singapore (1997) Somalia (1986)
South Africa (1997) Sri Lanka (1994) St. Vincent and the Grenadines (1994) Sudan (1972) Suriname (1993) Swaziland ((1994) Syria (1997) Togo (1981) Thailand (1994) Tonga (1994) Trinidad and Tobago (1994) Tunisia (1997) Turkey (1997) Taiwan (1997) Tanzania (1990) Uganda(1988) United Arab Emirates (1985) Uruguay(1997) Venezuela (1994) Western Samoa (1972) Yemen (1986) Zambia (1994) Zimbabwe (1997)
![[Global HR Forum 2014] Policies to Tackle Labour Market Duality in Korea : How Do Youth Start-Ups fit into the Big Picture?](https://img.pdfslide.us/doc/110x75/558cda23d8b42ad6118b4574/global-hr-forum-2014-policies-to-tackle-labour-market-duality-in-korea-how-do-youth-start-ups-fit-into-the-big-picture.jpg)
